• Military Menu                       

• Introduction
• Systems
• Facilities
• Agencies
• Industry
• Operations
• Countries
• Hot Documents
• News
• Reports
• Policy
• Budget
• Congress
• Links

• WMD
• Intelligence
• Homeland Security
• Space

NAVSEA Helps Validate Significant Advances in Facial Recognition Technology

NAVSEA News

By David Caskey, Naval Sea Systems Command Public Affairs

WASHINGTON -- Working with the Department of Commerce's National Institute of Standards (NIST), Naval Sea Systems Command (NAVSEA) has tested facial recognition systems and has found that the best applications are accurate about 90 percent of the time.

Computers have become much more adept at "recognizing" human faces during the past two years, and facial recognition technology has advanced to the point where automatic fingerprint matching systems were just five years ago.

Computer scientists from NAVSEA, NIST, the Defense Advanced Research Projects Agency (DARPA) and the Department of Defense's Counterdrug Technology Development Program completed a comprehensive evaluation of commercially available face recognition systems, and concluded that the technology has made significant advances. The study shows that there has been a 50 percent reduction in error rates since comparable tests were conducted in 2000.

Ten companies participated in the tests, which involved matching facial images provided by the U.S. Department of State from its consular office in Mexico. The tests involved matching 121,589 images of 37,437 individuals. The tests were administered at Naval Surface Warfare Center Dahlgren Division in Dahlgren, Va. from July 10-August 9, 2002.

Demographic results show that males are easier to identify than females, and older people are easier to recognize than younger people. A significant change in recognition also occurs over a three-year elapsed time between images.

The study also found significant differences in matching abilities depending on where the images are made. Face recognition in outdoor environments is only about half as good as images taken in indoor environments, where controlling lighting conditions is easier.

For verification (i.e. determining whether a person is who he or she claims to be) the best facial recognition systems are equivalent to 1998 fingerprint matching technologies, yielding a 90 percent probability of verification with a one percent probability of false acceptance.